Multi objective clustering for wireless sensor networks

Hacıoğlu G., Kand V. F. A., Sesli E.

EXPERT SYSTEMS WITH APPLICATIONS, vol.59, pp.86-100, 2016 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 59
  • Publication Date: 2016
  • Doi Number: 10.1016/j.eswa.2016.04.016
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.86-100
  • Keywords: Wireless sensor network, WSN, Clustering, nsga-2, Multi objective clustering, MULTIOBJECTIVE EVOLUTIONARY ALGORITHM
  • Karadeniz Technical University Affiliated: Yes


Ambient Intelligence is the next wave in computing and communication technology. Nano-sensors, wireless networks and unified intelligent software are the main elements of this issue. Inputs of ambient intelligence are taken from sensors in the environment. Wireless sensor networks consists of small and low cost sensors that collect and report environmental data. Wireless sensors are dispersed in an area that some phenomenon or event should be monitored. When a sensor detects the monitored event (heat, pressure, sound, light, areas having magnetic properties, vibration, etc.), the event is reported to one of the sites. This site performs an appropriate task such as sending a message or local processing based on the type of network, and then provides the appropriate response. One of the major challenges in wireless sensor networks is optimizing the energy consumption. Studies have shown that by clustering network nodes, it is possible to use their energy more efficiently. This study proposes a clustering based routing method to be used in wireless sensor networks. Multi-objective optimization algorithm named as Non dominated sorting genetic algorithm-II is used for clustering and seven objective functions are described. It is aimed to carry out several goals at once by using multi objective algorithm. While communication cost between the objective functions and cluster-head and Sink, and cluster-head and non-cluster-head is tried to be minimized, selection of the cluster heads only from the nodes near Sink is also tried to be prevented and it is also taken into consideration for clusters to include more nodes as much as possible. Each solution of the solution set obtained with Non-dominated sorting genetic algorithm-II points a different network topology. Each solution in solution set is the best solution according to some of objective functions. It is provided that Sink simulate each solution in solution set according to a certain scenario and choose one suitable for the desired criteria. In proposed method, both operating the Non-dominated sorting genetic algorithm-II and, simulation and evaluation of the obtained solutions comes out in Sink which has sufficient operation and power sources. According to the results, the proposed method can make the life span of network five times longer than LEACH, which is the most famous clustering algorithm. Besides, while the proposed method extends the life span of network, it is also seen that it increases the number of the packet reaching Sink two times more than LEACH. The data provided by proposed method includes information about larger areas when compared to LEACH. (C) 2016 Elsevier Ltd. All rights reserved.